Linear: The 5-Minute Owner's Manual

How the board fits your day — without becoming another inbox. Written July 31, 2026.

What I recommend

Treat Linear as your review-and-rank board, not a to-do list. Claude files the work and logs the executive calls; you spend 5 minutes a day approving or reversing, and 30 minutes on Monday re-ranking the week.

What you do + time cost

One 5-minute pass each morning — after the punch list, never before. One 30-minute pass on Monday. That's 55 minutes a week, total.

Why it matters

Your bottleneck audit found 12+ items stuck waiting on approvals nobody needed. The board is how decisions keep moving while you sell. If you don't review it, the discretion system has no brake; if you live in it, you're not making partner calls. Five minutes is the sweet spot.

First, the 60-second map of the board

You never need to build any of this — it's already set up. You just need to know what you're looking at.

Thing	What it means
5 columns	Inbox → Next → Doing → Waiting → Done. A card's column is the truth about where it stands, and it syncs both ways automatically.
4 projects	One per hero number: lead transfers to clients · qualified conversations from our own ads · booked calls and closes · cash collected. Every project is one bet on one number.
Parked project	The freezer. Everything that lost the 30-day focus fight lives here. You un-park by name, on purpose — it never thaws on its own.
Weekly cycles	Monday-start, one week long. A cycle is the week's commitment: what's in it is what we said we'd do.

The one rule that runs the whole board: a card that can't name a number is reactive. It doesn't get a project, doesn't get planned around, and gets batched when there's slack. If a card claims a project, it must move that project's number.

Technical notes (names and plumbing)

Team STP ("Stpierre"). ~58 cards are two-way synced to the Postgres brain via linear_issue_map; state changes flow Linear → brain through an n8n webhook receiver, and brain → Linear through a 5-minute push drain. State map: Inbox→inbox, Next→next, Doing→doing, Waiting→waiting, Done→done.

Projects, priority, cycles, and the reactive label are Linear-only view state — restructure them freely, nothing breaks.
Column moves and the agent-ready label DO sync. Renaming a workflow state or casually retitling/re-describing a card overwrites the brain copy — don't.
Never delete or cancel an issue. Additive and reversible only: dead work gets a label or a state move.

1 · The morning pass 5 min

Runs after the daily punch list on the command home page — never before. The punch list (read the page → answer every conversation → 10 partner calls → approve the ad wave → collect cash) is the money work. Linear is the check that nothing is silently stuck.

Open "Doing" and "Waiting" only. Ignore Inbox and the backlog entirely — mornings are not for triage.
Scan Waiting for your name. Anything waiting on you is the only thing that matters here. Unstick it: answer in a comment, or just do the thing. Target: Waiting-on-Akash = zero.
Glance at new decision cards (see section 2). Thumbs-up is silence — you only act on the ones you'd reverse.
Close the tab. If you're past 5 minutes, you're triaging. Stop — Monday owns triage.

What to ignore on purpose: Inbox, the Parked project, anything reactive, and card counts. A fat Inbox is not a problem in the morning. The only morning failure is something in Waiting that's been waiting on you for two days.

2 · Reviewing Claude's executive calls inside the 5 min

You told Claude: at a fork, decide and log it — don't queue it. That means the board carries a running record of calls already made. Your job is review, not approval. The work already happened.

What a logged call looks like

Switched the Dallas ad set to the schedule-optimized duplicate

The fork Old ad set optimizes for the wrong event; relight it or duplicate it.

The pick Duplicated with the right optimization, paused the old one.

The number Qualified conversations per week.

The undo Pause the duplicate, unpause the original. One click, reversible.

Every decision card carries those four lines: the fork · the pick · the number it serves · how to undo it. If a card is missing the undo line, that's a defect — comment "where's the undo?" and it gets fixed.

Your three moves, in order of force

Move	When	How
Silence	You'd have made the same call (this should be ~90% of them)	Do nothing. Silence is agreement — that's the whole point of the system.
Comment	The call stands, but the reasoning needs a correction for next time	One-line comment on the card. It becomes precedent without undoing work.
Reverse	Wrong call	Say "reverse [card]" in chat — chat is the reliable channel — and drop a comment on the card so the record shows why. The card's own undo line is the playbook; Claude executes it and the card reopens.

Never do this: quietly fix a wrong call yourself without touching the card. Then the record says the call was right, and the same fork gets decided the same wrong way next month. Reversals are cheap; silent drift is expensive.

Technical notes (what still requires an ask)

Claude's discretion stops at the Hard Rules list — these always come to you first, never as after-the-fact cards: money out, credential rotation, account permissions, deletions, outbound human messages sent as you, legal commitments, untested public launches. Everything else is decide-and-log. A standing "pending approval" state outside that list is drift — flag it.

3 · Adding work without flooding the board 30 sec / idea

The board almost died once: 22 of 28 cards piled into one project that named no number. The filter that fixed it: everything is non-essential by default. Only work you explicitly promote reaches the board — and only promoted board work ever reaches the agents.

The routing rule

It moves one of the five numbers this week (transfers · qualified conversations · booked calls · closes · cash) → say it to Claude or drop a card in Inbox. It gets a project, a priority, and a place in the cycle.
It's a real idea, just not this week → tell Claude "capture this." It goes to memory, not the board. Ideas are free; cards are commitments.
It's a maybe / someday → it goes to Parked or the park list, by name. It only comes back when you un-park it by name.

The test before any card gets a project: can you finish this sentence — "this moves [number] to [target] by [date]"? Can't finish it → it's reactive. It gets batched, never planned around. And no card sits at "No priority" — rank it when you file it or let Claude rank it.

Technical notes (how promotion actually works)

Every new card lands in the brain as lane='admin' (non-essential). Agents can only pick up a card when it's in the Next column and carries the agent-ready label — that combination flips it to lane='core' and writes a dispatch row. So handing work to the fleet from your phone is: drag to Next, add agent-ready. Nothing dispatches by accident. Add labels one at a time (the sync can drop labels on bulk label rewrites).

4 · The Monday weekly pass 30 min

Monday is when the board earns its keep. The cycle turns over Monday morning; this pass decides what the new week's commitment is. Do it in this order — retro before ranking, so last week's numbers pick this week's work.

Minutes 0–10 — Retro against the scorecard. Open the weekly scorecard next to Linear. For each hero number: predicted vs. actual. Where actual missed, find (or create) the "get it back on target" card — the miss must exist as a card, or it isn't real work.
Minutes 10–15 — Close out. Sweep Done. Confirm reversible decision cards you're at peace with. Anything in Doing that didn't move all week: demote it to Next or park it — don't let it squat.
Minutes 15–25 — Re-rank Next. Order by: (1) blocks a hero number, (2) hard external deadline, (3) how long it's been parked on you. The sprint goal — first St. Pierre cash collected by Aug 16 — is the tiebreaker on every close call.
Minutes 25–30 — Set the cycle. Pull this week's cards into the cycle. Fewer than you think — the cycle is a promise, not a wish list. Say the week's predicted numbers out loud to Claude so the scorecard has next Monday's targets.

One question ends every Monday pass: "If only the top three cards in Next happen this week, does cash still land by Aug 16?" If no, the ranking is wrong — fix it before you close the tab.

5 · What Linear is NOT for

Linear is the decision and commitment board — the record of what we chose and what we promised this week. Everything else already has a home. Putting it in Linear just makes two stale copies.

Don't put this in Linear	It lives here instead
Conversations	Chat with Claude, or Slack. A card comment is for the record, not for back-and-forth.
Documents & how-tos	The hosted doc pages. Cards link to docs; they don't contain them.
Memory & context	Claude's brain. Say "capture this" and it's remembered — no card needed.
Daily numbers	The command home page. The board tracks work, not metrics.
Every idea you have	Captured memory or the park list. A card is a commitment; ideas aren't commitments yet.
Client leads & sales convos	The CRM. Never mirror pipeline into cards.

Litmus test: "Will someone act on this card, then close it?" No → it's not a card.

Your first week 5 min/day

Don't try to learn the whole board at once. One habit per day; by Friday the 5-minute pass is muscle memory.

Day 1

Just look. Open the board after the punch list. Read the 4 project names and the 5 columns. Touch nothing. Close it.

Day 2

Read one decision card top to bottom — fork, pick, number, undo. If any of the four lines is missing, comment on it. That's your first review.

Day 3

Unstick one Waiting card. Find anything waiting on you, answer it in a comment or do it. Feel what "Waiting-on-Akash = zero" is like.

Day 4

File one card the right way. Take something from your head, drop it in Inbox with the "moves [number] to [target] by [date]" sentence. If you can't write the sentence, capture it to memory instead — that's the filter working.

Day 5

Promote one card to the agents. Drag a card to Next, add the agent-ready label, watch it get picked up. Now you can hand off work from your phone.

Day 6–7

Weekend: nothing. The board doesn't need you on weekends. Monday, run your first full 30-minute weekly pass with the scorecard open. That's the full system.

After week one: if the daily pass keeps taking more than 5 minutes, the board is too fat — that's a triage problem for Monday, not a reason to visit the board more often.